Multifractal analysis and lacunarity spectrum of the galaxies of the ninth Sloan Digital Sky Survey (SDSS) data release

Mon. Not. R. Astron. Soc. 000,[T](9](2012) Printed 21 December 2012 (MN KTbX style file v2.2)

Multifractal analysis and lacunarity spectrum of the galaxies of the
ninth Sloan Digital Sky Survey (SDSS) data release

C. A. Chacon-Cardona 1 ' 2 , R. A. Casas-Miranda 1

1 Departamento de Fi'sica, Universidad National de Colombia
2 Facultad Tecnoldgica, Universidad Distrital Francisco Jose de Caldas

Accepted — . Received — ; in original form —

ABSTRACT

In this work, we develop a statistical analysis of the large-scale clustering of matter in
the Universe from the fractal point of view using galaxies from the Ninth Sloan Digital Sky
Survey (SDSS) Data Release (DR9). From the total set of galaxies, a magnitude-limited sam-
ple of galaxies with redshifts in the range < z < 0.15 was created. The sample covers the
largest completely connected area of the celestial sphere within the catalogue, with limits in
right ascension of 120° < a < 240° and declination 0° < 5 < 60°, which is a region that
includes the largest galactic samples that have been studied from the fractal viewpoint to date.
The sample contains 164,168 galaxies.

Using the sliding-window technique, the multifractal dimension spectrum and its depen-
dence on radial distance are determined. This generalisation of the concept of fractal dimen-
sion is used to analyse large-scale clustering of matter in complex systems. Likewise, the
lacunarity spectrum, which is a quantity that complements the characterisation of a fractal set
by quantifying how the set fills the space in which it is embedded, is determined.

Using these statistical tools, we find that the clustering of galaxies exhibits fractal be-
haviour that depends on the radial distance for all calculated quantities. A transition to ho-
mogeneity is not observed in the calculation of the fractal dimension of galaxies; instead, the
galaxies exhibit a multifractal behaviour whose dimensional spectrum does not exceed the
physical spatial dimension for radial distances up to 180 Mpc/h from each centre within the
sample. Our results and their implications are discussed in the context of the formation of
large-scale structures in the Universe.

Key words: methods: statistical- galaxies -large-scale structure of Universe.

1 INTRODUCTION

The standard cosmological principle states that the cosmos is ho-
mogeneous and isotropic on large scales (Longair 20081, such that
the observed inhomogeneities are only perceived locally and should
vanish on sufficiently large scales, at distances on the order of hun-
dreds of Mpc/h l |Grujic & Pankovic|[2009l |Gabrielli et aT|2005) .
The majority of the developments in modern cosmology use the
cosmological principle as the fundamental hypothesis from which
the observations and theoretical developments are contrasted and
interpreted. The success of the solution to the Einstein field equa-
tions for a universe consisting of an ideal gas of galaxies, found by
Friedmann, Robertson and Walker, makes the corresponding met-
ric the representation of the space-time structure of the Universe
that is largely accepted in cosmology for the study of large-scale
matter clustering. The ACDM model strongly relies on the stan-
dard cosmological principle because it introduces the cosmological
constant as a remedy for the problem of the acceleration of the Uni-

verse, which is confirmed by observations of Type la supernovae.
However, the model excludes possible solutions that do not assume
homogeneity, such as models based on the metric of Tolman and
Bondi l |Ribeiro| 1 992) |Enqvist|2008) .

Despite the success of physical models based on the stan-
dard cosmological principle, there are still many unresolved funda-
mental questions regarding the formation of structure in the Uni-
verse (Peacock 1999). For example, is the cosmological princi-
ple supported by the observations of galaxies in the most recent
catalogues? If so, then what is the distance scale at which the
transition to homogeneity is observed? Some research groups re-
port that galaxies and other astrophysical objects, such as quasars,
are grouped in highly structured hierarchical patterns that exhibit
properties of self-similarity (scale invariance) and a fractal dimen-
sion smaller than the physical space dimension (Rozgacheva et al.
|2012| l. Similar results are found for the anisotropies of the cos-
mic background radiation (Kobayashi et al. 201 1 1. In such works,
a revision of the cosmological principle is proposed because, in a

2 C. A. Chacon- Cardona, R. A. Casas-Miranda

universe with fractal clustering of galaxies, models based on ho-
mogeneity and isotropy would not be consistent with observations.
Therefore, new methods of interpreting galaxy clustering informa-
tion from the perspective of statistical mechanics should be devel-
oped ( (Gabrielli et al.||2005| >. Furthermore, other research groups
(e.g., ([Sarkar et a l.|2009^ ) claim to have found the scale at which the
cosmological principle is valid. In their multifractal analysis of the
Sixth SDSS Data Release (DR6), they find a transition to homo-
geneity at approximately 70 Mpc/h. Similarly, analyses of galac-
tic catalogues with information from the southern hemisphere con-
clude using a method based on the correlation dimension - that the
transition to homogeneity occurs on the same spatial scale ( |Scrim-|
|geour et al.|2012) .

In this article, we present calculations of the multifractal di-
mension and the corresponding lacunarity spectrum based on the
generalised correlation integral for all galaxies in the Ninth SDSS
Data Release (DR9). We use the same methodology that was ap-
plied to dark matter halos from the Millennium Simulation, an N-
body simulation with gravitational interaction. For the simulated
halos, we found a transition to homogeneity at radial distances be-
tween 100 Mpc/h and 120 Mpc/h, and the lacunarity spectrum con-
firmed such transition ( Chacon-Cardona & Casas-Miranda 2012).

This paper is organised as follows. In section 2, the mathe-
matical foundations involved in the determination of multifractal-
ity and the lacunarity spectrum of galaxy clustering is reviewed.
Section 3 contains a summary of the characteristics of DR9 and
of the primary magnitude-limited sample of galaxies that we use.
In section 4, the calculations of multifractality and lacunarity are
discussed for the case of galaxy clustering up to a redshift of 0.15,
which corresponds to a maximum radial distance of approximately
600 Mpc/h, where the effect of space-time curvature is negligible.
In section 5, the results of multifractality and lacunarity are dis-
cussed, and the conclusions are presented in section 6.

the subset of chosen centres. Here rii(r), is the number of particles
within a radial distance r with respect to a particle i chosen as the
centre. This expression is defined as:

rii{r) = 6( r —

I).

(2)

i=i

where the sum is taken over all the particles in the sample. The
coordinates of each particle in our three-dimensional space are de-
noted by Xj , and O is the Heaviside function, which is defined as
Q(x) = for x < and 0(a;) = 1 for x ^ 0. The number of par-
ticles around each centre is determined by counting the number of
galaxies around the centre, i.e., the galaxies found within a sphere
of radius r around the centre.

From Equation l[TJ, the correlation dimension is expressed in
the following form:

C 2 (r) = F/

(3)

where F is called the pre-factor of the power law and is the correla-
tion dimension of the set under study. Thus, the correlation dimen-
sion is calculated as the derivative:

Do =

d log C 2 (r)
d Ioe r

(4)

By expressing the correlation integral as a power law, in ad-
dition to the multifractal dimension, the set is fully characterised
by the proportionality constant F that accompanies the function.
The pre-factor F which is related to the average distance between
the nearest neighbours, permits defining the lacunarity, which is a
property that determines how the fractal set fills the space. Defined
by |Blumenfel d & Mandelbrot| (T997) via the second-order variabil-
ity factor for the pre-factor F, the lacunarity is expressed as:

2 THEORETICAL FORMALISM

2.1 Correlation Dimension and Lacunarity

Fractal analysis of the clustering of galaxies is based on the idea of
generalising the dimension of a metric space. This concept, devel-
oped by Felix Hausdorff in ( Hausdorff 1918), allowed Benoit Man-
delbrot to define a fractal as a set with Hausdorff dimension strictly
exceeding its topological dimension ((Mandelbrot 1983 1, where the
topological dimension is the number of independent directions that
can be taken around an element belonging to a given set. It is possi-
ble to determine the Hausdorff dimension for finite sets using com-
putational algorithms, such as the minimum spanning tree (MST).
However, the correlation dimension D2, is a very useful tool for
measuring fractality on large scales (Martinez & Saari 2002| >.

The definition of the correlation dimension is based on the in-
tegral of the correlation function C2, which is a function that mea-
sures the number of neighbours found on average within a certain
radius r around a given centre. In agreement with the notation in
the literature used by Bagla, Yadav & Seshadri (2008 1, it is defined
as:

((F-(F)f) _ <F 2 )

(FY

(FY

(5)

Whereas lacunarity was defined as a complementary concept
to the fractal mass-radius dimension, this concept can be applied
to other definitions of fractal dimension, as shown in (Mar tinez &|
|Saar|20 02); The correlation dimension and its corresponding lacu-
narity are sufficient to characterise sets for which these quantities
do not depend on the scale, as discussed in the majority of studies
that demonstrate fractal behaviour, at least on small scales. There-
fore, in the case of scale dependence of fractal quantities we must
extend the statistical analysis by using the formalism of the mul-
tifractal dimension (Nakamichi & Morikawa 2004; Gabrielli et al.
[20051 ).

2.2 Generalised Fractal Dimension

Based on the correlation integral C2(r), given Equation {T|, the
generalised correlation integral can be defined via the relation:

(1)

where TV is the total number of particles within the distribution, M
is the number of particles used as centres, and the sum is taken over

1 M

19-1

(6)

where M is the number of centres, TV is the number of particles
included in the sample, rii(r) is the same expression defined in
Equation |2](, and q is the structure parameter, which corresponds

Multifractal and lacunarity analysis of the galaxies ofSDSS Data Release 9 3

to an arbitrary real number. From the generalised correlation in-
tegral, it is possible to make a power series expansion of log(r)
as described in Provenzale, Spiegel & Thieberger ( 19971, and thus
be able to calculate the multifractal dimension and the lacunarity
spectrum directly:

log [C q {r) 1/{q - 1) ] = D q log{r) + log{F q )

+ o

(7)

By retaining only the first two terms on the right side of the equa-
tion, we obtain the relationship between the generalised correlation
integral and the multifractal dimension:

C q {r)

1/(9-1)

(8)

The generalised fractal dimension and the generalised lacu-
narity can be defined in the same manner by which the authors of
|Blumenfeld & M andelbrot ( 1997), defined these concepts for the
mass-radius fractal relation:

D q =

d\ogC q (r

(9-1) d\

(9)

Likewise, the corresponding generalised lacunarity based on
the pre-factor F q for each structure parameter for q is defined by:

((F q -(F q )f) (F* q )

(10)

In the fractal dimension spectrum, if holds for any q ^ q',
D q — D q i then the distribution is said to be a homogeneous fractal
(monofractal). q 1, D q characterises the scaling behaviour of the
system in high-density environments, i.e., clusters or super-clusters
of galaxies. Conversely, for values q < 1, characterises the scaling
behaviour in low-density environments, i.e., vacuum regions ( |Gaite|
|2007|[Sarkar et al.|2009| l. When the distribution of galaxies exceeds
the transition to homogeneity, all the values of the multifractal di-
mension should approach the physical dimension of the space and
the lacunarity spectrum must approach zero on the same radial dis-
tance scale r.

Next, we apply the concepts of generalised fractal dimension
and lacunarity to the distribution of galaxies from the latest update
of the SDSS catalogue.

3 GALAXIES FROM THE NINTH DATA RELEASE OF
THE SDSS CATALOGUE

The SDSS galactic catalogue is the largest astronomical database
accumulated by humankind to date. Its ninth edition contains infor-
mation regarding at least 1231,051,050 astrophysical objects, in-
cluding spectral information for 668,054 stars, 1,457,002 galaxies
and 228,468 quasars and their respective locations in the celestial
sphere over an area of 14,555 square degrees (SDSS -III Collabf>|
|ration et al.|2012} . In its ninth data release, the catalogue has two
clearly defined regions without any information gaps in their inte-
riors, therefore in these regions it is possible to calculate the fractal
dimension without the risk of losing accuracy. From these two re-
gions, we chose the region with the largest connected area to have
a more representative sample of the galaxy catalogue; the region
covers a total area of 7,200 square degrees.

The information contained in this data release has been the

■

160 180 200

Right Ascension a[°]

Figure 1. Distribution on the celestial sphere of galaxies located in the
largest area completely covered by SDSS. The region is bounded by equa-
torial coordinates of right ascension 120° < a < 240° and declination

0° < <5 < 60°

subject of two different targeting algorithms, one for galaxies with
redshifts in the range 0.15 < z < 0.4, called LOWZ, and a sec-
ond sample of galaxies at higher redshifts, with0.4 < z < 0.8 ,
called CMASS. While we could use either of these two samples
for the fractal characterisation of galaxy clustering, we wanted to
avoid erroneously introducing homogenising effects arising from
the determination of distances according to the standard cosmo-
logical model, as pointed by Ribeiro ( Ribeiro 2005 ) and the group
(Gabrielli et al. 2005 1. Using Hubble's law, which is an experimen-
tal fact valid for astrophysical object localisation at low redshift
(z <C 1), we selected from DR9 galaxies with redshifts in the range
2.3 x 10~ 3 < z < 0.15, corresponding to distances greater than 10
Mpc/h, where the effect of peculiar velocities is small, and less than
600 Mpc/h, where the difference between the distance determined
using the Hubble law and the co-moving distance determined ac-
cording to the standard cosmological model does not exceed 4%..
In this region, the targeting algorithm most suitable for reliable de-
tection of galaxies is the Main Sample of galaxies, a Petrosian r-
band magnitude-limited galactic sample, whose selection process
is described in detail by Strauss et al. (2002)

3.1 Main sample of galaxies

The SDSS redshift catalogue includes galaxy photometric data cor-
rected for galactic extinction, where reddening calculations are per-
formed according to the maps of infrared emission from galactic
dust described in Schlegel, Finkbeiner & Davis ( 1998). All infor-
mation necessary for fractal analysis is provided by the catalogue
in CASJOBS, which is accessed using SQL (Syntax Query Lan-
guage). First, galaxy candidates are separated from stars. Then, two
cuts in apparent magnitude are applied for the Petrosian r spectral
band, which is located at approximately 6165 . The apparent mag-
nitude for acceptable targets is in the interval 15 < m r < 17.77,
where the first limit indicates that those targets with apparent mag-
nitude less than 15 are discarded, i.e., very bright objects that cause
contamination in the sample are discarded. We also discard objects
with Petrosian radii, which contain half the total flux of the galaxy,
greater than two arcseconds (r p > 2"). The second limit indicates
that it is not possible to detect objects that are fainter in apparent
magnitude than the limit of 17.77. In addition, it is necessary to

4 C. A. Chacon- Cardona, R. A. Casas-Miranda

-15 ™ 1 1 1 1 1 r

Magnitude Limited Sample

Mi

-16 -

-i7 - llSs,

-22 I I I I I I I

100 200 300 400 500 600 700

Hubble Distance r[Mpc/hl

Figure 2. Magnitude-radial distance diagram for the primary sample of
galaxies from DR9 used for the multifractal calculations. The sample is
composed of 164,168 galaxies with redshifts in the range 2.3 X 10 -3 <
z < 0.15.

apply a condition on the average surface brightness determined ac-
cording to the Petrosian radius; the most significant cut is the cut
that includes targets with values /iso ^ 23mag / arcsec 2 .

Given the apparent magnitude limits, to obtain a sample that
is not incomplete because of the effects of selection by brightness,
the procedure includes the determination of the volume that lim-
its the sample according to the range of distances over which one
wishes to work. Because our interest is measuring fractality with-
out introducing any a priori cosmological model, with distances in
the range lOMpc/h ^ r ^ QQQMpc/h the absolute magnitude of
the galaxies in the sample are in the range —15 ^ M ^ —21.5, as
shown in Figure|2]

Finally, for the volume-limited sample, we must determine
the radius of the largest sphere in which fractality calculations are
performed without introducing assumptions about the shape of the
clustering to be analysed. The effective depth scale for our sample
according to the limits in right ascension 120° < a < 240°, and
declination 0° < S < 60°, is determined in accord with (Gabrielli
|et al.|20 05 ) by the relationship:

_ R d sin(S6/2)
s l + «n(tf0/2)' ( '

where Rd corresponds to the maximum radial distance of the sam-
ple and <5l9 is

89 = minia-i — ati, Si — 5i) (12)

In our case the maximum radius is R s — 202 Mpc/h.

Starting with this sample, the next step involves the determi-
nation of the multifractal dimension and lacunarity spectrum up to
depths of 202 Mpc/h, which corresponds to the effective depth scale
of our volume-limited sample that relies on a number of available
centres for the analysis of 81,435 galaxies

4 MULTIFRACTALITY AND LACUNARITY SPECTRUM
FOR GALAXY CLUSTERING IN SDSS DR9

In the previous section, we analysed the volume-limited sample of
galaxies with redshifts in the range 2.3 x 10~ 3 < z < 0.15 .

This set is composed of 164,168 galaxies for which we calculated
the generalised correlation integral. We used 13 values of the struc-
ture parameter in the range — 6 ^ q ^ 6 in steps of 1, which are
presented in Figure [3] These calculations include high-density re-
gions, i.e., (q 1), and low-density regions (q < 1). For q = 1,
the numerical limit is shown in the figure.

The fractal dimension spectrum is determined from the gener-
alised correlation integral using the sliding-window technique. This
technique is applied to the logarithm of the correlation integral as
a function of the logarithm of the radial distance. This function is
approximated by straight lines using the least squares method. Be-
low, Figure [4] shows the results of the calculations of the fractal
dimension spectrum for low-density regions with structure param-
eter values q < 1.

For high-density regions, namely for the correlation integral
for structure parameter values ??in the range q ^ 1, the depen-
dence of the spectrum of multifractal dimension on radial distance
is shown in Figure [5]

These results can be combined to determine the multifractal
dimension spectrum as a function of the structure parameter q for
different radial distance scales, as shown in Figure|6]

Given that the fractal dimension does not specify how galax-
ies fill space, our analysis is completed by the calculation of the
lacunarity spectrum for the same values of the structure parameter.
Figure|7]shows the corresponding results.

5 DISCUSSION

Chacon-Cardona & Casas-Miranda (2012) have performed fractal
analysis on the clustering of dark matter halos from the Millen-
nium Simulation (Springel et al. 2005). We successfully applied
the same methodology to the clustering of galaxies in DR9 for the
local Universe, i.e., galaxies for which the Hubble distance can
be used for the localisation of points in the set without using an
a priori cosmological model. The analysed sample was obtained
from the most complete and connected region within the galac-
tic catalogue (with right ascension in the range 120 < a < 240
and declination in the range < S < 60 ). The sample is a
magnitude-limited sample of 164,168 galaxies with redshifts in
the range 2.3 x 10~ 3 < z < 0.15. The calculations performed
accounted for the effective depth for which the fractal behaviour
could be analysed, up to a maximum radial distance of 202 Mpc/h,
which is the maximum radius of the sphere that we can inscribe
within the limits of our sample of galaxies without introducing bias
for subsequent statistical analysis (Gabrielli et al. 2005}.

We directly calculated the multifractal dimension using the
correlation integral. The generalised correlation integral found for
galaxy clustering behaves as expected and is in agreement with the
results of (Chacon-Cardona & Casas-Miranda 2 012| > and (Sarkar
|et al.|20 09l. The resulting behaviour corresponds to an increase in
the average number of neighbours that are located around the cen-
tres in high-density regions, for which q ^ 1, and a decrease in the
number of neighbours for low-density regions (voids with q < 1).

Using the sliding-window technique, the fractal dimension
was calculated for low-density regions with structural parameter
values q < 1. In agreement with the findings for the Millennium
Simulation (Chacon-Cardona & Casas-Miranda 2012), we found a
growth that surpasses the physical dimension of space in regions
near the centres, followed by a rapid tendency toward homogene-
ity at radial distances greater than 50 Mpc/h. In contrast, for high-
density regions with structure parameter values q 1, the be-

Multifractal and lacunarity analysis of the galaxies ofSDSS Data Release 9 5

0.1 I 10 100 1000 0.1 1 10 100 1000

r(Mpc/h) r(Mpc/h)

Figure 3. Generalised correlation integral for all values of q studied in this article for our volume-limited sample from DR9. Densities with q ^ 1 appear to
the left, whereas low densities, with q < 1, appear on the right. All the graphs are shown in log scale

haviour differs from that we found previously ( Chacor TCardona &]
|Casas- Miranda 2012). Although there is a region of rapid dimen-
sional growth, asymptotic behaviour toward the physical dimension
of space is not observed. In our case, the dimension reaches a high
value in regions around the centres, where the tendency towards
homogeneity is observed, followed by a second region where the
dimension of the galaxy clustering tends to differ from the physi-
cal dimension of space, thus showing persistent fractal behaviour
at larger scales. This is consistent with galaxy clustering behaviour
on large scales, as published by other research groups Joyc e et aT|
(2005 l, [Verevkin, Bukhmastova & Baryshev| ( |2011}, It is also con-
sistent with analysis of the clustering of quasars [Rozgachev a et al.|
( |2012} and with studies of the anisotropy of the cosmic background
radiation Kobayashi et al. (201 1 1.

The determination of the lacunarity spectrum is spoiled by
the propagation of errors inherent in its calculation based on the
pre-factor F. Notwithstanding this difficulty, oscillatory behaviour
(voids alternating with regions of high concentration of galaxies)
can be detected, as was found in the Millennium Simulation, with
a lacunarity maximum in regions below 10 Mpc/h. However, in
our case, monotonically decreasing behaviour was not observed.
This result is consistent with deviations from the transition to ho-
mogeneity detected by the multifractal-dimension set because the
lacunarity values increase again after 50 Mpc/h.

6 CONCLUSIONS

In this article, we analysed galaxy clustering in the local Universe
from the multifractal point of view. Our results led us to conclude
the following:

No transition to homogeneity was observed in galaxy cluster-
ing, at least for distance scales less than approximately 180 Mpc/h;
instead, there was a clear persistence of multifractal behaviour. For
values of the structure parameter corresponding to low densities,
the calculated dimensions exhibited an unequivocal tendency to fill
space almost homogeneously at scales greater than 50 Mpc/h. How-
ever, for galaxy clustering, the fractal-dimension spectrum in all
cases showed two clearly defined regions: a first region of high-
dimensional growth on scales less than 30 Mpc/h and a second
region of smoothly decaying structure with fractal dimension be-
low that of physical space at depths greater than approximately 30

Mpc/h. Therefore, the use of multifractal analysis is most suitable
for the characterisation of galaxy clustering.

The lacunarity spectrum confirmed the persistence of the frac-
tal behaviour of the set, showing oscillatory behaviour with alter-
nation between high and low lacunarity values, i.e., the presence
of voids followed by regions of high galaxy clustering. We did not
observe a lacunarity decrease that would otherwise allow detecting
homogeneity. This result confirms the departure from homogeneity
within the analysed volume.

The sample is eminently multifractal. A tendency towards
monofractality (i.e., only one fractal-dimension value independent
of the structure parameter) was not observed in the galactic sample
for the distance scale of this sample. This result implies that fur-
ther detailed research is needed to investigate the standard cosmo-
logical principle and its consistency with observations. Although
we cannot conclude that a transition to homogeneity cannot occur
at greater depths, the problem of introducing homogenisation due
to cosmological assumptions should be avoided, with the goal of
approaching this problem without introducing biases in the FRW
metric.

ACKNOWLEDGEMENTS

C. A. Chacon-Cardona thanks the Universidad Distrital Francisco
Jose de Caldas for their financial support to perform doctoral stud-
ies.

REFERENCES

Bagla J. S., Yadav J., Seshadri T. R., 2008, MNRAS, 390, 829
Blumenfeld R., Mandelbrot B. B., 1997, Phys. Rev. E, 56, 1 12
Chacon-Cardona C. A., Casas-Miranda R. A., 2012, MNRAS,
427, 2613

Enqvist K., 2008, General Relativity and Gravitation, 40, 451
Gabrielli A., Sylos, Joyce M., Pietronero L., 2005, Statistical

Physics for Cosmic Structures. Springer Verlag
Gaite J., 2007, ApJ, 658, 11
Grujic P., Pankovic V., 2009, ArXiv e-prints
Hausdorff F, 1918, Mathematische Annalen, 79, 157,

10.1007/BF01457179

6 C. A. Chacdn-Cardona, R. A. Casas-Miranda

Multifractal Dimension D. t
Physical Space Dimension

Multifractal Dimension D 5 h
Physical Space Dimension -

r(M|ic/li>

r(Mpc/h)

Multifractal Dimension D 4 h
Physical Space Dimension ■

100

r(Mpc/h)

Multifractal Dimension D 2 h
Physical Space Dimension

100

r(Mpc/h)

8
7
6
5

Mulli fractal Dimension D i
Physical Space Dimension

Multifractal Dimension D 3 r
Physical Space Dimension ■

100

r(Mpe/h)

MultitVactal Dimension D , i
Physical Space Dimension -

100

r(Mpc/h)

r(Mpc/li)

Figure 4. Multi-fractal spectrum D q (r) as a function of the radial distance r for low-density environments, with ?6 ^ q ^ 0, corresponding to the galaxy
clustering in SDSS DR9. Error bars ±lcr are indicated in the graphs.

Multifractal and lacunarity analysis of the galaxies ofSDSS Data Release 9 7

r

MuhiiVactal Dimension D, t
Physical Space Dimension

r(Mpc/h)

Multifractal Dimension D, h
Physical Space Dimension

LOO
r(Mpc/h)

Multifractal Dimension D^ t
Physical Space Dimension

Hit)
r(Mpc/h)

Multifractal Dimension D, t
Physical Space Dimension -

r(Mpc/h)

Multifractal Dimension D 4 h
Physical Space Dimension -

11X1

r(Mpc/h)

Multifractal Dimension D (l
Physical Space Dimension

100

r(Mpc/h)

Figure 5. Multifractal spectrum D q (r) as a function of the radial distance r for low-density environments, with —6 ^ q ^ 0, corresponding to the galaxy
clustering in SDSS DR9. Error bars, corresponding to ±1<t are indicated in the plots.

Joyce M., Sylos Labini R, Gabrielli A., Montuori M., Pietronero
L., 2005, A&A, 443, 11

Kobayashi N., Yamazaki Y., Kuninaka H., Katori M., Matsushita
M., Matsushita S., Chiang L.-Y, 2011, Journal of the Physical
Society of Japan, 80, 074003

Longair M., 2008, Galaxy formation, Astronomy and astrophysics
library. Springer

Mandelbrot B., 1983, The fractal geometry of nature. W.H. Free-
man

Martinez V., Saar E., 2002, Statistics of the galaxy distribution.
Chapman & Hall/CRC

Nakamichi A., Morikawa M., 2004, Physica A Statistical Me-
chanics and its Applications, 341, 215

Peacock J., 1999, Cosmological physics. Cambridge University

Press

Provenzale A., Spiegel E. A., Thieberger R., 1997, Chaos, 7, 82
Ribeiro M. B„ 1992, ApJ, 388, 1
— , 2005, A&A, 429, 65

Rozgacheva I. K., Borisov A. A., Agapov A. A., Pozdneev I. A.,

Shchetinina O. A., 2012, ArXiv e-prints
Sarkar P., Yadav J., Pandey B., Bharadwaj S., 2009, MNRAS, 399,

L128

Schlegel D. J., Finkbeiner D. P., Davis M., 1998, ApJ, 500, 525
Scrimgeour M., Davis T., Blake C, James J. B., Poole G., et al.,

2012, Mon.Not.Roy.Astron.Soc, 425, 116
SDSS-III Collaboration et al., 2012, ArXiv e-prints
Springel V. et al., 2005, Nature, 435, 629
Strauss M. A. et al, 2002, AJ, 124, 1810

Figure 6. Multi-fractal dimension spectrum D q (r) of the galaxy clustering as a function of the structure parameter q in the range [6,6]. Multi-fractal behaviour
is observed even at depths on the order of 180 Mpc/h around the centres. The solid line indicates the Bezier interpolation curve.

Multifractal and lacunarity analysis of the galaxies ofSDSS Data Release 9 9

* 0.0001

r(Mpc/h)

Multifractal Lacunarity 4> 3 i

100 150
R(Mpc/h)

<J 0.0001

^ 0.001

1000
100
10

1

0.1
0.01
0.001
0.0001
le-05
le-06

100

150

R(Mpc/h)

Multifractal Lacunarity (f

Multifractal Lacunarity <ti 6

50 100 150

R(Mpc/h)

Figure 7. Lacunarity spectrum & q (r) in losgscale as function of the radial distance r for high-density environments, with q 1, according to SDSS (DR9).
Error bars of ±l<x are indicated in the graphs.

Verevkin A. O., Bukhmastova Y. L., Baryshev Y. V., 2011, As-
tronomy Reports, 55, 324